|
1 |
/* -*- C++ -*-
|
|
2 |
*
|
|
3 |
* This file is a part of LEMON, a generic C++ optimization library
|
|
4 |
*
|
|
5 |
* Copyright (C) 2003-2008
|
|
6 |
* Egervary Jeno Kombinatorikus Optimalizalasi Kutatocsoport
|
|
7 |
* (Egervary Research Group on Combinatorial Optimization, EGRES).
|
|
8 |
*
|
|
9 |
* Permission to use, modify and distribute this software is granted
|
|
10 |
* provided that this copyright notice appears in all copies. For
|
|
11 |
* precise terms see the accompanying LICENSE file.
|
|
12 |
*
|
|
13 |
* This software is provided "AS IS" with no warranty of any kind,
|
|
14 |
* express or implied, and with no claim as to its suitability for any
|
|
15 |
* purpose.
|
|
16 |
*
|
|
17 |
*/
|
|
18 |
|
|
19 |
#ifndef LEMON_KARP_H
|
|
20 |
#define LEMON_KARP_H
|
|
21 |
|
|
22 |
/// \ingroup shortest_path
|
|
23 |
///
|
|
24 |
/// \file
|
|
25 |
/// \brief Karp's algorithm for finding a minimum mean cycle.
|
|
26 |
|
|
27 |
#include <vector>
|
|
28 |
#include <limits>
|
|
29 |
#include <lemon/core.h>
|
|
30 |
#include <lemon/path.h>
|
|
31 |
#include <lemon/tolerance.h>
|
|
32 |
#include <lemon/connectivity.h>
|
|
33 |
|
|
34 |
namespace lemon {
|
|
35 |
|
|
36 |
/// \brief Default traits class of Karp algorithm.
|
|
37 |
///
|
|
38 |
/// Default traits class of Karp algorithm.
|
|
39 |
/// \tparam GR The type of the digraph.
|
|
40 |
/// \tparam LEN The type of the length map.
|
|
41 |
/// It must conform to the \ref concepts::ReadMap "ReadMap" concept.
|
|
42 |
#ifdef DOXYGEN
|
|
43 |
template <typename GR, typename LEN>
|
|
44 |
#else
|
|
45 |
template <typename GR, typename LEN,
|
|
46 |
bool integer = std::numeric_limits<typename LEN::Value>::is_integer>
|
|
47 |
#endif
|
|
48 |
struct KarpDefaultTraits
|
|
49 |
{
|
|
50 |
/// The type of the digraph
|
|
51 |
typedef GR Digraph;
|
|
52 |
/// The type of the length map
|
|
53 |
typedef LEN LengthMap;
|
|
54 |
/// The type of the arc lengths
|
|
55 |
typedef typename LengthMap::Value Value;
|
|
56 |
|
|
57 |
/// \brief The large value type used for internal computations
|
|
58 |
///
|
|
59 |
/// The large value type used for internal computations.
|
|
60 |
/// It is \c long \c long if the \c Value type is integer,
|
|
61 |
/// otherwise it is \c double.
|
|
62 |
/// \c Value must be convertible to \c LargeValue.
|
|
63 |
typedef double LargeValue;
|
|
64 |
|
|
65 |
/// The tolerance type used for internal computations
|
|
66 |
typedef lemon::Tolerance<LargeValue> Tolerance;
|
|
67 |
|
|
68 |
/// \brief The path type of the found cycles
|
|
69 |
///
|
|
70 |
/// The path type of the found cycles.
|
|
71 |
/// It must conform to the \ref lemon::concepts::Path "Path" concept
|
|
72 |
/// and it must have an \c addBack() function.
|
|
73 |
typedef lemon::Path<Digraph> Path;
|
|
74 |
};
|
|
75 |
|
|
76 |
// Default traits class for integer value types
|
|
77 |
template <typename GR, typename LEN>
|
|
78 |
struct KarpDefaultTraits<GR, LEN, true>
|
|
79 |
{
|
|
80 |
typedef GR Digraph;
|
|
81 |
typedef LEN LengthMap;
|
|
82 |
typedef typename LengthMap::Value Value;
|
|
83 |
#ifdef LEMON_HAVE_LONG_LONG
|
|
84 |
typedef long long LargeValue;
|
|
85 |
#else
|
|
86 |
typedef long LargeValue;
|
|
87 |
#endif
|
|
88 |
typedef lemon::Tolerance<LargeValue> Tolerance;
|
|
89 |
typedef lemon::Path<Digraph> Path;
|
|
90 |
};
|
|
91 |
|
|
92 |
|
|
93 |
/// \addtogroup shortest_path
|
|
94 |
/// @{
|
|
95 |
|
|
96 |
/// \brief Implementation of Karp's algorithm for finding a minimum
|
|
97 |
/// mean cycle.
|
|
98 |
///
|
|
99 |
/// This class implements Karp's algorithm for finding a directed
|
|
100 |
/// cycle of minimum mean length (cost) in a digraph.
|
|
101 |
///
|
|
102 |
/// \tparam GR The type of the digraph the algorithm runs on.
|
|
103 |
/// \tparam LEN The type of the length map. The default
|
|
104 |
/// map type is \ref concepts::Digraph::ArcMap "GR::ArcMap<int>".
|
|
105 |
#ifdef DOXYGEN
|
|
106 |
template <typename GR, typename LEN, typename TR>
|
|
107 |
#else
|
|
108 |
template < typename GR,
|
|
109 |
typename LEN = typename GR::template ArcMap<int>,
|
|
110 |
typename TR = KarpDefaultTraits<GR, LEN> >
|
|
111 |
#endif
|
|
112 |
class Karp
|
|
113 |
{
|
|
114 |
public:
|
|
115 |
|
|
116 |
/// The type of the digraph
|
|
117 |
typedef typename TR::Digraph Digraph;
|
|
118 |
/// The type of the length map
|
|
119 |
typedef typename TR::LengthMap LengthMap;
|
|
120 |
/// The type of the arc lengths
|
|
121 |
typedef typename TR::Value Value;
|
|
122 |
|
|
123 |
/// \brief The large value type
|
|
124 |
///
|
|
125 |
/// The large value type used for internal computations.
|
|
126 |
/// Using the \ref KarpDefaultTraits "default traits class",
|
|
127 |
/// it is \c long \c long if the \c Value type is integer,
|
|
128 |
/// otherwise it is \c double.
|
|
129 |
typedef typename TR::LargeValue LargeValue;
|
|
130 |
|
|
131 |
/// The tolerance type
|
|
132 |
typedef typename TR::Tolerance Tolerance;
|
|
133 |
|
|
134 |
/// \brief The path type of the found cycles
|
|
135 |
///
|
|
136 |
/// The path type of the found cycles.
|
|
137 |
/// Using the \ref KarpDefaultTraits "default traits class",
|
|
138 |
/// it is \ref lemon::Path "Path<Digraph>".
|
|
139 |
typedef typename TR::Path Path;
|
|
140 |
|
|
141 |
/// The \ref KarpDefaultTraits "traits class" of the algorithm
|
|
142 |
typedef TR Traits;
|
|
143 |
|
|
144 |
private:
|
|
145 |
|
|
146 |
TEMPLATE_DIGRAPH_TYPEDEFS(Digraph);
|
|
147 |
|
|
148 |
// Data sturcture for path data
|
|
149 |
struct PathData
|
|
150 |
{
|
|
151 |
bool found;
|
|
152 |
LargeValue dist;
|
|
153 |
Arc pred;
|
|
154 |
PathData(bool f = false, LargeValue d = 0, Arc p = INVALID) :
|
|
155 |
found(f), dist(d), pred(p) {}
|
|
156 |
};
|
|
157 |
|
|
158 |
typedef typename Digraph::template NodeMap<std::vector<PathData> >
|
|
159 |
PathDataNodeMap;
|
|
160 |
|
|
161 |
private:
|
|
162 |
|
|
163 |
// The digraph the algorithm runs on
|
|
164 |
const Digraph &_gr;
|
|
165 |
// The length of the arcs
|
|
166 |
const LengthMap &_length;
|
|
167 |
|
|
168 |
// Data for storing the strongly connected components
|
|
169 |
int _comp_num;
|
|
170 |
typename Digraph::template NodeMap<int> _comp;
|
|
171 |
std::vector<std::vector<Node> > _comp_nodes;
|
|
172 |
std::vector<Node>* _nodes;
|
|
173 |
typename Digraph::template NodeMap<std::vector<Arc> > _out_arcs;
|
|
174 |
|
|
175 |
// Data for the found cycle
|
|
176 |
LargeValue _cycle_length;
|
|
177 |
int _cycle_size;
|
|
178 |
Node _cycle_node;
|
|
179 |
|
|
180 |
Path *_cycle_path;
|
|
181 |
bool _local_path;
|
|
182 |
|
|
183 |
// Node map for storing path data
|
|
184 |
PathDataNodeMap _data;
|
|
185 |
// The processed nodes in the last round
|
|
186 |
std::vector<Node> _process;
|
|
187 |
|
|
188 |
Tolerance _tolerance;
|
|
189 |
|
|
190 |
public:
|
|
191 |
|
|
192 |
/// \name Named Template Parameters
|
|
193 |
/// @{
|
|
194 |
|
|
195 |
template <typename T>
|
|
196 |
struct SetLargeValueTraits : public Traits {
|
|
197 |
typedef T LargeValue;
|
|
198 |
typedef lemon::Tolerance<T> Tolerance;
|
|
199 |
};
|
|
200 |
|
|
201 |
/// \brief \ref named-templ-param "Named parameter" for setting
|
|
202 |
/// \c LargeValue type.
|
|
203 |
///
|
|
204 |
/// \ref named-templ-param "Named parameter" for setting \c LargeValue
|
|
205 |
/// type. It is used for internal computations in the algorithm.
|
|
206 |
template <typename T>
|
|
207 |
struct SetLargeValue
|
|
208 |
: public Karp<GR, LEN, SetLargeValueTraits<T> > {
|
|
209 |
typedef Karp<GR, LEN, SetLargeValueTraits<T> > Create;
|
|
210 |
};
|
|
211 |
|
|
212 |
template <typename T>
|
|
213 |
struct SetPathTraits : public Traits {
|
|
214 |
typedef T Path;
|
|
215 |
};
|
|
216 |
|
|
217 |
/// \brief \ref named-templ-param "Named parameter" for setting
|
|
218 |
/// \c %Path type.
|
|
219 |
///
|
|
220 |
/// \ref named-templ-param "Named parameter" for setting the \c %Path
|
|
221 |
/// type of the found cycles.
|
|
222 |
/// It must conform to the \ref lemon::concepts::Path "Path" concept
|
|
223 |
/// and it must have an \c addFront() function.
|
|
224 |
template <typename T>
|
|
225 |
struct SetPath
|
|
226 |
: public Karp<GR, LEN, SetPathTraits<T> > {
|
|
227 |
typedef Karp<GR, LEN, SetPathTraits<T> > Create;
|
|
228 |
};
|
|
229 |
|
|
230 |
/// @}
|
|
231 |
|
|
232 |
public:
|
|
233 |
|
|
234 |
/// \brief Constructor.
|
|
235 |
///
|
|
236 |
/// The constructor of the class.
|
|
237 |
///
|
|
238 |
/// \param digraph The digraph the algorithm runs on.
|
|
239 |
/// \param length The lengths (costs) of the arcs.
|
|
240 |
Karp( const Digraph &digraph,
|
|
241 |
const LengthMap &length ) :
|
|
242 |
_gr(digraph), _length(length), _comp(digraph), _out_arcs(digraph),
|
|
243 |
_cycle_length(0), _cycle_size(1), _cycle_node(INVALID),
|
|
244 |
_cycle_path(NULL), _local_path(false), _data(digraph)
|
|
245 |
{}
|
|
246 |
|
|
247 |
/// Destructor.
|
|
248 |
~Karp() {
|
|
249 |
if (_local_path) delete _cycle_path;
|
|
250 |
}
|
|
251 |
|
|
252 |
/// \brief Set the path structure for storing the found cycle.
|
|
253 |
///
|
|
254 |
/// This function sets an external path structure for storing the
|
|
255 |
/// found cycle.
|
|
256 |
///
|
|
257 |
/// If you don't call this function before calling \ref run() or
|
|
258 |
/// \ref findMinMean(), it will allocate a local \ref Path "path"
|
|
259 |
/// structure. The destuctor deallocates this automatically
|
|
260 |
/// allocated object, of course.
|
|
261 |
///
|
|
262 |
/// \note The algorithm calls only the \ref lemon::Path::addFront()
|
|
263 |
/// "addFront()" function of the given path structure.
|
|
264 |
///
|
|
265 |
/// \return <tt>(*this)</tt>
|
|
266 |
Karp& cycle(Path &path) {
|
|
267 |
if (_local_path) {
|
|
268 |
delete _cycle_path;
|
|
269 |
_local_path = false;
|
|
270 |
}
|
|
271 |
_cycle_path = &path;
|
|
272 |
return *this;
|
|
273 |
}
|
|
274 |
|
|
275 |
/// \name Execution control
|
|
276 |
/// The simplest way to execute the algorithm is to call the \ref run()
|
|
277 |
/// function.\n
|
|
278 |
/// If you only need the minimum mean length, you may call
|
|
279 |
/// \ref findMinMean().
|
|
280 |
|
|
281 |
/// @{
|
|
282 |
|
|
283 |
/// \brief Run the algorithm.
|
|
284 |
///
|
|
285 |
/// This function runs the algorithm.
|
|
286 |
/// It can be called more than once (e.g. if the underlying digraph
|
|
287 |
/// and/or the arc lengths have been modified).
|
|
288 |
///
|
|
289 |
/// \return \c true if a directed cycle exists in the digraph.
|
|
290 |
///
|
|
291 |
/// \note <tt>mmc.run()</tt> is just a shortcut of the following code.
|
|
292 |
/// \code
|
|
293 |
/// return mmc.findMinMean() && mmc.findCycle();
|
|
294 |
/// \endcode
|
|
295 |
bool run() {
|
|
296 |
return findMinMean() && findCycle();
|
|
297 |
}
|
|
298 |
|
|
299 |
/// \brief Find the minimum cycle mean.
|
|
300 |
///
|
|
301 |
/// This function finds the minimum mean length of the directed
|
|
302 |
/// cycles in the digraph.
|
|
303 |
///
|
|
304 |
/// \return \c true if a directed cycle exists in the digraph.
|
|
305 |
bool findMinMean() {
|
|
306 |
// Initialization and find strongly connected components
|
|
307 |
init();
|
|
308 |
findComponents();
|
|
309 |
|
|
310 |
// Find the minimum cycle mean in the components
|
|
311 |
for (int comp = 0; comp < _comp_num; ++comp) {
|
|
312 |
if (!initComponent(comp)) continue;
|
|
313 |
processRounds();
|
|
314 |
updateMinMean();
|
|
315 |
}
|
|
316 |
return (_cycle_node != INVALID);
|
|
317 |
}
|
|
318 |
|
|
319 |
/// \brief Find a minimum mean directed cycle.
|
|
320 |
///
|
|
321 |
/// This function finds a directed cycle of minimum mean length
|
|
322 |
/// in the digraph using the data computed by findMinMean().
|
|
323 |
///
|
|
324 |
/// \return \c true if a directed cycle exists in the digraph.
|
|
325 |
///
|
|
326 |
/// \pre \ref findMinMean() must be called before using this function.
|
|
327 |
bool findCycle() {
|
|
328 |
if (_cycle_node == INVALID) return false;
|
|
329 |
IntNodeMap reached(_gr, -1);
|
|
330 |
int r = _data[_cycle_node].size();
|
|
331 |
Node u = _cycle_node;
|
|
332 |
while (reached[u] < 0) {
|
|
333 |
reached[u] = --r;
|
|
334 |
u = _gr.source(_data[u][r].pred);
|
|
335 |
}
|
|
336 |
r = reached[u];
|
|
337 |
Arc e = _data[u][r].pred;
|
|
338 |
_cycle_path->addFront(e);
|
|
339 |
_cycle_length = _length[e];
|
|
340 |
_cycle_size = 1;
|
|
341 |
Node v;
|
|
342 |
while ((v = _gr.source(e)) != u) {
|
|
343 |
e = _data[v][--r].pred;
|
|
344 |
_cycle_path->addFront(e);
|
|
345 |
_cycle_length += _length[e];
|
|
346 |
++_cycle_size;
|
|
347 |
}
|
|
348 |
return true;
|
|
349 |
}
|
|
350 |
|
|
351 |
/// @}
|
|
352 |
|
|
353 |
/// \name Query Functions
|
|
354 |
/// The results of the algorithm can be obtained using these
|
|
355 |
/// functions.\n
|
|
356 |
/// The algorithm should be executed before using them.
|
|
357 |
|
|
358 |
/// @{
|
|
359 |
|
|
360 |
/// \brief Return the total length of the found cycle.
|
|
361 |
///
|
|
362 |
/// This function returns the total length of the found cycle.
|
|
363 |
///
|
|
364 |
/// \pre \ref run() or \ref findMinMean() must be called before
|
|
365 |
/// using this function.
|
|
366 |
LargeValue cycleLength() const {
|
|
367 |
return _cycle_length;
|
|
368 |
}
|
|
369 |
|
|
370 |
/// \brief Return the number of arcs on the found cycle.
|
|
371 |
///
|
|
372 |
/// This function returns the number of arcs on the found cycle.
|
|
373 |
///
|
|
374 |
/// \pre \ref run() or \ref findMinMean() must be called before
|
|
375 |
/// using this function.
|
|
376 |
int cycleArcNum() const {
|
|
377 |
return _cycle_size;
|
|
378 |
}
|
|
379 |
|
|
380 |
/// \brief Return the mean length of the found cycle.
|
|
381 |
///
|
|
382 |
/// This function returns the mean length of the found cycle.
|
|
383 |
///
|
|
384 |
/// \note <tt>alg.cycleMean()</tt> is just a shortcut of the
|
|
385 |
/// following code.
|
|
386 |
/// \code
|
|
387 |
/// return static_cast<double>(alg.cycleLength()) / alg.cycleArcNum();
|
|
388 |
/// \endcode
|
|
389 |
///
|
|
390 |
/// \pre \ref run() or \ref findMinMean() must be called before
|
|
391 |
/// using this function.
|
|
392 |
double cycleMean() const {
|
|
393 |
return static_cast<double>(_cycle_length) / _cycle_size;
|
|
394 |
}
|
|
395 |
|
|
396 |
/// \brief Return the found cycle.
|
|
397 |
///
|
|
398 |
/// This function returns a const reference to the path structure
|
|
399 |
/// storing the found cycle.
|
|
400 |
///
|
|
401 |
/// \pre \ref run() or \ref findCycle() must be called before using
|
|
402 |
/// this function.
|
|
403 |
const Path& cycle() const {
|
|
404 |
return *_cycle_path;
|
|
405 |
}
|
|
406 |
|
|
407 |
///@}
|
|
408 |
|
|
409 |
private:
|
|
410 |
|
|
411 |
// Initialization
|
|
412 |
void init() {
|
|
413 |
if (!_cycle_path) {
|
|
414 |
_local_path = true;
|
|
415 |
_cycle_path = new Path;
|
|
416 |
}
|
|
417 |
_cycle_path->clear();
|
|
418 |
_cycle_length = 0;
|
|
419 |
_cycle_size = 1;
|
|
420 |
_cycle_node = INVALID;
|
|
421 |
for (NodeIt u(_gr); u != INVALID; ++u)
|
|
422 |
_data[u].clear();
|
|
423 |
}
|
|
424 |
|
|
425 |
// Find strongly connected components and initialize _comp_nodes
|
|
426 |
// and _out_arcs
|
|
427 |
void findComponents() {
|
|
428 |
_comp_num = stronglyConnectedComponents(_gr, _comp);
|
|
429 |
_comp_nodes.resize(_comp_num);
|
|
430 |
if (_comp_num == 1) {
|
|
431 |
_comp_nodes[0].clear();
|
|
432 |
for (NodeIt n(_gr); n != INVALID; ++n) {
|
|
433 |
_comp_nodes[0].push_back(n);
|
|
434 |
_out_arcs[n].clear();
|
|
435 |
for (OutArcIt a(_gr, n); a != INVALID; ++a) {
|
|
436 |
_out_arcs[n].push_back(a);
|
|
437 |
}
|
|
438 |
}
|
|
439 |
} else {
|
|
440 |
for (int i = 0; i < _comp_num; ++i)
|
|
441 |
_comp_nodes[i].clear();
|
|
442 |
for (NodeIt n(_gr); n != INVALID; ++n) {
|
|
443 |
int k = _comp[n];
|
|
444 |
_comp_nodes[k].push_back(n);
|
|
445 |
_out_arcs[n].clear();
|
|
446 |
for (OutArcIt a(_gr, n); a != INVALID; ++a) {
|
|
447 |
if (_comp[_gr.target(a)] == k) _out_arcs[n].push_back(a);
|
|
448 |
}
|
|
449 |
}
|
|
450 |
}
|
|
451 |
}
|
|
452 |
|
|
453 |
// Initialize path data for the current component
|
|
454 |
bool initComponent(int comp) {
|
|
455 |
_nodes = &(_comp_nodes[comp]);
|
|
456 |
int n = _nodes->size();
|
|
457 |
if (n < 1 || (n == 1 && _out_arcs[(*_nodes)[0]].size() == 0)) {
|
|
458 |
return false;
|
|
459 |
}
|
|
460 |
for (int i = 0; i < n; ++i) {
|
|
461 |
_data[(*_nodes)[i]].resize(n + 1);
|
|
462 |
}
|
|
463 |
return true;
|
|
464 |
}
|
|
465 |
|
|
466 |
// Process all rounds of computing path data for the current component.
|
|
467 |
// _data[v][k] is the length of a shortest directed walk from the root
|
|
468 |
// node to node v containing exactly k arcs.
|
|
469 |
void processRounds() {
|
|
470 |
Node start = (*_nodes)[0];
|
|
471 |
_data[start][0] = PathData(true, 0);
|
|
472 |
_process.clear();
|
|
473 |
_process.push_back(start);
|
|
474 |
|
|
475 |
int k, n = _nodes->size();
|
|
476 |
for (k = 1; k <= n && int(_process.size()) < n; ++k) {
|
|
477 |
processNextBuildRound(k);
|
|
478 |
}
|
|
479 |
for ( ; k <= n; ++k) {
|
|
480 |
processNextFullRound(k);
|
|
481 |
}
|
|
482 |
}
|
|
483 |
|
|
484 |
// Process one round and rebuild _process
|
|
485 |
void processNextBuildRound(int k) {
|
|
486 |
std::vector<Node> next;
|
|
487 |
Node u, v;
|
|
488 |
Arc e;
|
|
489 |
LargeValue d;
|
|
490 |
for (int i = 0; i < int(_process.size()); ++i) {
|
|
491 |
u = _process[i];
|
|
492 |
for (int j = 0; j < int(_out_arcs[u].size()); ++j) {
|
|
493 |
e = _out_arcs[u][j];
|
|
494 |
v = _gr.target(e);
|
|
495 |
d = _data[u][k-1].dist + _length[e];
|
|
496 |
if (!_data[v][k].found) {
|
|
497 |
next.push_back(v);
|
|
498 |
_data[v][k] = PathData(true, _data[u][k-1].dist + _length[e], e);
|
|
499 |
}
|
|
500 |
else if (_tolerance.less(d, _data[v][k].dist)) {
|
|
501 |
_data[v][k] = PathData(true, d, e);
|
|
502 |
}
|
|
503 |
}
|
|
504 |
}
|
|
505 |
_process.swap(next);
|
|
506 |
}
|
|
507 |
|
|
508 |
// Process one round using _nodes instead of _process
|
|
509 |
void processNextFullRound(int k) {
|
|
510 |
Node u, v;
|
|
511 |
Arc e;
|
|
512 |
LargeValue d;
|
|
513 |
for (int i = 0; i < int(_nodes->size()); ++i) {
|
|
514 |
u = (*_nodes)[i];
|
|
515 |
for (int j = 0; j < int(_out_arcs[u].size()); ++j) {
|
|
516 |
e = _out_arcs[u][j];
|
|
517 |
v = _gr.target(e);
|
|
518 |
d = _data[u][k-1].dist + _length[e];
|
|
519 |
if (!_data[v][k].found || _tolerance.less(d, _data[v][k].dist)) {
|
|
520 |
_data[v][k] = PathData(true, d, e);
|
|
521 |
}
|
|
522 |
}
|
|
523 |
}
|
|
524 |
}
|
|
525 |
|
|
526 |
// Update the minimum cycle mean
|
|
527 |
void updateMinMean() {
|
|
528 |
int n = _nodes->size();
|
|
529 |
for (int i = 0; i < n; ++i) {
|
|
530 |
Node u = (*_nodes)[i];
|
|
531 |
if (!_data[u][n].found) continue;
|
|
532 |
LargeValue length, max_length = 0;
|
|
533 |
int size, max_size = 1;
|
|
534 |
bool found_curr = false;
|
|
535 |
for (int k = 0; k < n; ++k) {
|
|
536 |
if (!_data[u][k].found) continue;
|
|
537 |
length = _data[u][n].dist - _data[u][k].dist;
|
|
538 |
size = n - k;
|
|
539 |
if (!found_curr || length * max_size > max_length * size) {
|
|
540 |
found_curr = true;
|
|
541 |
max_length = length;
|
|
542 |
max_size = size;
|
|
543 |
}
|
|
544 |
}
|
|
545 |
if ( found_curr && (_cycle_node == INVALID ||
|
|
546 |
max_length * _cycle_size < _cycle_length * max_size) ) {
|
|
547 |
_cycle_length = max_length;
|
|
548 |
_cycle_size = max_size;
|
|
549 |
_cycle_node = u;
|
|
550 |
}
|
|
551 |
}
|
|
552 |
}
|
|
553 |
|
|
554 |
}; //class Karp
|
|
555 |
|
|
556 |
///@}
|
|
557 |
|
|
558 |
} //namespace lemon
|
|
559 |
|
|
560 |
#endif //LEMON_KARP_H
|